This long-running grant deals with the persistent language impairment that results from left hemisphere stroke (chronic aphasia). Over the years, it produced new aphasia assessment tools, experimental methods, and computational and statistical techniques, and made these available to the research community. Unique to our approach is the emphasis on linking data to implemented computational models of individual patients and patient variability that are derived from psycholinguistic and cognitive principles. These data and models now form the backbone of lesion-mapping and learning studies that are the focus of the next cycle. Building on our success in applying group-level, voxel-based lesion-symptom mapping (VLSM) methodology, the next cycle will introduce new enhancements in the form of multivariate lesion-behavior mapping at the level of voxels, regions, and tracks. Coupling these powerful techniques to our model-driven behavioral analysis ensures that the "symptoms" that are mapped are realistic with respect to psycholinguistic models, and meaningful with respect to patient performance. Specific aims are to identify lesion correlates of impairments at conceptual, lexical-semantic, and lexical-phonological stages of word production. There is consensus on the need to bridge the gap between theory and practice in language rehabilitation. Towards this end, a series of studies are proposed that aim to articulate principles of learning in connection with the learning-based recovery of lexical- semantic and lexical-phonological access to words in aphasia. As in our past work, the innovation is the linking of individual patients and patient variability to theory and models. Specifically, we strive to formulate principles of learning that specify how impairment in the lexical access system affects the system's ability to monitor its own state, strengthen weights through repetition, learn from errors, and so on. Such principles constitute building blocks for effective, individually tailored treatment strategies. They are building blocks, too, for learning-based approaches to recovery of other, more complex language functions. PUBLIC HEALTH RELEVANCE: Our research is relevant to the diagnosis of acquired disorders of language, the mapping of language functions in the human brain, and the development of theoretically motivated rehabilitation strategies. An ongoing commitment is to provide ready access by others to the methods and data developed under this project.